In [9]:
img = Image.open('/content/archive/test/angry/im106.png')
img.show()
Team Members: Virajkumar Tank (101411542)
Problem Statement: I will try to fine-tune a model by changing its parameters, implementing pre-trained model in order to achive higher accuracy and also I will prepare it for emotion detection in video.
Dataset Description: I am going to develop a deep learning neural network model which will detect Emotions of Humans. The dataset contain 35,685 examples of 48x48 pixel gray scale images of faces divided into train and test dataset. The training set includes 28,709 images, the testing set includes 7178 images. Images are categorized based on the emotion shown in the facial expressions (happiness, neutral, sadness, anger, surprise, disgust, fear).
The existing notebook I have used to classify level of water bottle has accuracy of around 60-65% and Convolutional Neural Network(CNN) is being used but accuracy of the model is not upto mark.
Kaggle Dataset Link: https://www.kaggle.com/datasets/ananthu017/emotion-detection-fer/code?resource=download
Existing notebook link: https://www.kaggle.com/code/aryan348/emotion-detection-cnn
I tried to increase the accuracy by fine tunning of CNN model and also tried to capture emotions from a video.
import os
import random
import zipfile
from PIL import Image
import numpy as np # linear algebra
import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)
import matplotlib.pyplot as plt
import seaborn as sns
import tensorflow as tf
import keras
from keras.preprocessing import image
from keras.models import Sequential
from keras.layers import Conv2D, MaxPool2D, Flatten,Dense,Dropout,BatchNormalization
from tensorflow.keras.preprocessing.image import ImageDataGenerator
from tensorflow.keras.models import load_model
from tensorflow.python.keras import regularizers
from tensorflow.keras.preprocessing.image import load_img, img_to_array
from keras.preprocessing import image
# Unzip the downloaded file
zip_ref = zipfile.ZipFile("archive.zip", "r")
zip_ref.extractall()
zip_ref.close()
dataset_path = "/content/archive"
labels = os.listdir(dataset_path)
print(labels)
for label in labels:
print(os.listdir(os.path.join(dataset_path, label))) #listing number of folders and sub-folders in dataset
[]
for dirname, _, filenames in os.walk('/kaggle/input'):
for filename in filenames:
print(os.path.join(dirname, filename))
# Input data files are available in the read-only "../input/" directory
# For example, running this (by clicking run or pressing Shift+Enter) will list all files under the input directory
train_dir = "/content/archive/train"
test_dir = "../content/archive/test"
#preparing training data
img_size=48
train_datagen = ImageDataGenerator( width_shift_range = 0.1,
height_shift_range = 0.1,
horizontal_flip = True,
rescale = 1./255,
validation_split = 0.2
)
train_generator = train_datagen.flow_from_directory(directory = train_dir,
target_size = (img_size,img_size),
batch_size = 64,
color_mode = "grayscale",
class_mode = "categorical",
subset = "training"
)
Found 22968 images belonging to 7 classes.
#preparing validation dataset
validation_datagen = ImageDataGenerator(rescale = 1./255,
validation_split = 0.2)
validation_generator = validation_datagen.flow_from_directory( directory = test_dir,
target_size = (img_size,img_size),
batch_size = 64,
color_mode = "grayscale",
class_mode = "categorical",
subset = "validation"
)
Found 1432 images belonging to 7 classes.
img = Image.open('/content/archive/test/angry/im106.png')
img.show()
image in grayscale
img = load_img('/content/archive/test/angry/im106.png', target_size=(48, 48), color_mode = "grayscale")
img = np.array(img)
plt.imshow(img)
print(img.shape)
(48, 48)
train_ds = tf.keras.preprocessing.image_dataset_from_directory(
directory='/content/archive',
validation_split=0.2,
subset="training",
seed=123,
image_size=(224, 224),
batch_size=32)
Found 35887 files belonging to 2 classes. Using 28710 files for training.
# Get a random sample from the training dataset
random_sample = train_ds.take(1)
# Display the images in the random sample
import matplotlib.pyplot as plt
plt.figure(figsize=(10, 10))
for images, labels in random_sample:
for i in range(8):
ax = plt.subplot(3, 3, i + 1)
plt.imshow(images[i].numpy().astype("uint8"))
plt.axis("off")
from tensorflow.keras.optimizers import Adam,RMSprop,SGD,Adamax
model1= tf.keras.models.Sequential()
model1.add(Conv2D(32, kernel_size=(3, 3), padding='same', activation='relu', input_shape=(48, 48,1)))
model1.add(Conv2D(64,(3,3), padding='same', activation='relu' ))
model1.add(BatchNormalization())
model1.add(MaxPool2D(pool_size=(2, 2)))
model1.add(Dropout(0.25))
model1.add(Conv2D(128,(5,5), padding='same', activation='relu'))
model1.add(BatchNormalization())
model1.add(MaxPool2D(pool_size=(2, 2)))
model1.add(Dropout(0.25))
model1.add(Conv2D(512,(3,3), padding='same', activation='relu', kernel_regularizer=regularizers.l2(0.01)))
model1.add(BatchNormalization())
model1.add(MaxPool2D(pool_size=(2, 2)))
model1.add(Dropout(0.25))
model1.add(Flatten())
model1.add(Dense(256,activation = 'relu'))
model1.add(BatchNormalization())
model1.add(Dropout(0.25))
model1.add(Dense(512,activation = 'relu'))
model1.add(BatchNormalization())
model1.add(Dropout(0.25))
model1.add(Dense(7, activation='softmax'))
model1.compile(
optimizer = Adam(lr=0.0001),
loss='categorical_crossentropy',
metrics=['accuracy']
)
WARNING:absl:`lr` is deprecated, please use `learning_rate` instead, or use the legacy optimizer, e.g.,tf.keras.optimizers.legacy.Adam.
model1.summary()
Model: "sequential"
_________________________________________________________________
Layer (type) Output Shape Param #
=================================================================
conv2d (Conv2D) (None, 48, 48, 32) 320
conv2d_1 (Conv2D) (None, 48, 48, 64) 18496
batch_normalization (BatchN (None, 48, 48, 64) 256
ormalization)
max_pooling2d (MaxPooling2D (None, 24, 24, 64) 0
)
dropout (Dropout) (None, 24, 24, 64) 0
conv2d_2 (Conv2D) (None, 24, 24, 128) 204928
batch_normalization_1 (Batc (None, 24, 24, 128) 512
hNormalization)
max_pooling2d_1 (MaxPooling (None, 12, 12, 128) 0
2D)
dropout_1 (Dropout) (None, 12, 12, 128) 0
conv2d_3 (Conv2D) (None, 12, 12, 512) 590336
batch_normalization_2 (Batc (None, 12, 12, 512) 2048
hNormalization)
max_pooling2d_2 (MaxPooling (None, 6, 6, 512) 0
2D)
dropout_2 (Dropout) (None, 6, 6, 512) 0
flatten (Flatten) (None, 18432) 0
dense (Dense) (None, 256) 4718848
batch_normalization_3 (Batc (None, 256) 1024
hNormalization)
dropout_3 (Dropout) (None, 256) 0
dense_1 (Dense) (None, 512) 131584
batch_normalization_4 (Batc (None, 512) 2048
hNormalization)
dropout_4 (Dropout) (None, 512) 0
dense_2 (Dense) (None, 7) 3591
=================================================================
Total params: 5,673,991
Trainable params: 5,671,047
Non-trainable params: 2,944
_________________________________________________________________
epochs= 50
batch_size=60
history = model1.fit(x = train_generator,epochs = epochs,validation_data = validation_generator)
Epoch 1/50 359/359 [==============================] - 45s 84ms/step - loss: 3.0613 - accuracy: 0.2375 - val_loss: 2.7137 - val_accuracy: 0.1425 Epoch 2/50 359/359 [==============================] - 27s 74ms/step - loss: 1.9217 - accuracy: 0.3173 - val_loss: 1.9814 - val_accuracy: 0.3003 Epoch 3/50 359/359 [==============================] - 28s 77ms/step - loss: 1.8640 - accuracy: 0.3929 - val_loss: 2.3227 - val_accuracy: 0.3073 Epoch 4/50 359/359 [==============================] - 27s 74ms/step - loss: 1.7673 - accuracy: 0.4407 - val_loss: 1.7744 - val_accuracy: 0.4595 Epoch 5/50 359/359 [==============================] - 27s 75ms/step - loss: 1.7228 - accuracy: 0.4742 - val_loss: 1.5790 - val_accuracy: 0.5126 Epoch 6/50 359/359 [==============================] - 27s 74ms/step - loss: 1.6823 - accuracy: 0.4899 - val_loss: 1.6165 - val_accuracy: 0.5126 Epoch 7/50 359/359 [==============================] - 27s 76ms/step - loss: 1.6302 - accuracy: 0.5133 - val_loss: 1.7272 - val_accuracy: 0.5307 Epoch 8/50 359/359 [==============================] - 26s 74ms/step - loss: 1.6899 - accuracy: 0.5162 - val_loss: 1.7970 - val_accuracy: 0.5168 Epoch 9/50 359/359 [==============================] - 29s 81ms/step - loss: 1.6892 - accuracy: 0.5160 - val_loss: 1.8801 - val_accuracy: 0.4986 Epoch 10/50 359/359 [==============================] - 27s 74ms/step - loss: 1.6137 - accuracy: 0.5300 - val_loss: 1.5849 - val_accuracy: 0.5454 Epoch 11/50 359/359 [==============================] - 28s 77ms/step - loss: 1.5796 - accuracy: 0.5289 - val_loss: 1.4619 - val_accuracy: 0.5622 Epoch 12/50 359/359 [==============================] - 27s 74ms/step - loss: 1.5773 - accuracy: 0.5418 - val_loss: 1.6526 - val_accuracy: 0.5363 Epoch 13/50 359/359 [==============================] - 28s 77ms/step - loss: 1.5818 - accuracy: 0.5434 - val_loss: 1.5680 - val_accuracy: 0.5845 Epoch 14/50 359/359 [==============================] - 27s 75ms/step - loss: 1.5641 - accuracy: 0.5495 - val_loss: 1.5250 - val_accuracy: 0.5517 Epoch 15/50 359/359 [==============================] - 27s 75ms/step - loss: 1.5639 - accuracy: 0.5517 - val_loss: 1.5351 - val_accuracy: 0.5663 Epoch 16/50 359/359 [==============================] - 28s 77ms/step - loss: 1.5936 - accuracy: 0.5546 - val_loss: 2.0013 - val_accuracy: 0.4504 Epoch 17/50 359/359 [==============================] - 27s 75ms/step - loss: 1.6094 - accuracy: 0.5541 - val_loss: 1.5561 - val_accuracy: 0.5698 Epoch 18/50 359/359 [==============================] - 29s 80ms/step - loss: 1.5292 - accuracy: 0.5563 - val_loss: 1.4997 - val_accuracy: 0.5719 Epoch 19/50 359/359 [==============================] - 27s 75ms/step - loss: 1.5246 - accuracy: 0.5625 - val_loss: 1.5688 - val_accuracy: 0.5594 Epoch 20/50 359/359 [==============================] - 28s 77ms/step - loss: 1.5013 - accuracy: 0.5678 - val_loss: 1.5455 - val_accuracy: 0.5733 Epoch 21/50 359/359 [==============================] - 27s 74ms/step - loss: 1.5036 - accuracy: 0.5731 - val_loss: 1.4990 - val_accuracy: 0.5852 Epoch 22/50 359/359 [==============================] - 28s 77ms/step - loss: 1.5069 - accuracy: 0.5773 - val_loss: 1.4098 - val_accuracy: 0.5817 Epoch 23/50 359/359 [==============================] - 27s 75ms/step - loss: 1.4676 - accuracy: 0.5777 - val_loss: 1.4438 - val_accuracy: 0.5943 Epoch 24/50 359/359 [==============================] - 28s 77ms/step - loss: 1.5625 - accuracy: 0.5813 - val_loss: 1.4516 - val_accuracy: 0.5950 Epoch 25/50 359/359 [==============================] - 26s 74ms/step - loss: 1.5540 - accuracy: 0.5762 - val_loss: 1.5253 - val_accuracy: 0.5775 Epoch 26/50 359/359 [==============================] - 27s 76ms/step - loss: 1.5502 - accuracy: 0.5822 - val_loss: 1.5528 - val_accuracy: 0.5803 Epoch 27/50 359/359 [==============================] - 27s 74ms/step - loss: 1.5142 - accuracy: 0.5785 - val_loss: 1.5269 - val_accuracy: 0.5684 Epoch 28/50 359/359 [==============================] - 28s 77ms/step - loss: 1.5665 - accuracy: 0.5799 - val_loss: 1.5397 - val_accuracy: 0.5880 Epoch 29/50 359/359 [==============================] - 26s 74ms/step - loss: 1.4770 - accuracy: 0.5894 - val_loss: 1.4856 - val_accuracy: 0.5845 Epoch 30/50 359/359 [==============================] - 28s 77ms/step - loss: 1.4584 - accuracy: 0.5903 - val_loss: 1.5343 - val_accuracy: 0.5733 Epoch 31/50 359/359 [==============================] - 27s 75ms/step - loss: 1.5127 - accuracy: 0.5898 - val_loss: 1.4891 - val_accuracy: 0.6034 Epoch 32/50 359/359 [==============================] - 28s 77ms/step - loss: 1.4879 - accuracy: 0.5975 - val_loss: 1.5268 - val_accuracy: 0.5943 Epoch 33/50 359/359 [==============================] - 27s 76ms/step - loss: 1.4249 - accuracy: 0.5958 - val_loss: 1.3238 - val_accuracy: 0.6159 Epoch 34/50 359/359 [==============================] - 28s 77ms/step - loss: 1.3916 - accuracy: 0.5997 - val_loss: 1.4273 - val_accuracy: 0.5936 Epoch 35/50 359/359 [==============================] - 27s 74ms/step - loss: 1.3995 - accuracy: 0.5962 - val_loss: 1.3895 - val_accuracy: 0.5964 Epoch 36/50 359/359 [==============================] - 28s 77ms/step - loss: 1.4610 - accuracy: 0.6007 - val_loss: 1.4285 - val_accuracy: 0.6089 Epoch 37/50 359/359 [==============================] - 27s 74ms/step - loss: 1.3971 - accuracy: 0.6002 - val_loss: 1.4999 - val_accuracy: 0.5915 Epoch 38/50 359/359 [==============================] - 28s 77ms/step - loss: 1.4293 - accuracy: 0.6031 - val_loss: 1.4832 - val_accuracy: 0.6047 Epoch 39/50 359/359 [==============================] - 27s 74ms/step - loss: 1.3788 - accuracy: 0.6071 - val_loss: 1.4653 - val_accuracy: 0.6061 Epoch 40/50 359/359 [==============================] - 28s 77ms/step - loss: 1.3906 - accuracy: 0.6062 - val_loss: 1.4458 - val_accuracy: 0.5957 Epoch 41/50 359/359 [==============================] - 27s 74ms/step - loss: 1.3779 - accuracy: 0.6098 - val_loss: 1.3423 - val_accuracy: 0.6020 Epoch 42/50 359/359 [==============================] - 29s 81ms/step - loss: 1.3476 - accuracy: 0.6113 - val_loss: 1.3532 - val_accuracy: 0.5943 Epoch 43/50 359/359 [==============================] - 36s 102ms/step - loss: 1.3615 - accuracy: 0.6142 - val_loss: 1.3910 - val_accuracy: 0.6020 Epoch 44/50 359/359 [==============================] - 27s 74ms/step - loss: 1.3469 - accuracy: 0.6138 - val_loss: 1.5481 - val_accuracy: 0.6159 Epoch 45/50 359/359 [==============================] - 27s 76ms/step - loss: 1.4966 - accuracy: 0.6078 - val_loss: 1.3670 - val_accuracy: 0.6082 Epoch 46/50 359/359 [==============================] - 26s 74ms/step - loss: 1.4755 - accuracy: 0.6089 - val_loss: 1.4709 - val_accuracy: 0.6110 Epoch 47/50 359/359 [==============================] - 28s 77ms/step - loss: 1.3357 - accuracy: 0.6167 - val_loss: 1.3384 - val_accuracy: 0.6034 Epoch 48/50 359/359 [==============================] - 26s 73ms/step - loss: 1.3152 - accuracy: 0.6198 - val_loss: 1.3498 - val_accuracy: 0.6187 Epoch 49/50 359/359 [==============================] - 27s 76ms/step - loss: 1.3051 - accuracy: 0.6192 - val_loss: 1.3272 - val_accuracy: 0.6208 Epoch 50/50 359/359 [==============================] - 27s 76ms/step - loss: 1.3475 - accuracy: 0.6189 - val_loss: 1.3427 - val_accuracy: 0.6264
# get the list of all file paths in train_ds
file_paths = train_ds.file_paths
# randomly select 10 file paths
random_files = random.sample(file_paths, 10)
# load the images corresponding to the random file paths
images = []
for file_path in random_files:
img = tf.keras.preprocessing.image.load_img(file_path, target_size=(224, 224))
img_array = tf.keras.preprocessing.image.img_to_array(img)
images.append(img_array)
fig , ax = plt.subplots(1,2)
train_acc = history.history['accuracy']
train_loss = history.history['loss']
fig.set_size_inches(12,4)
ax[0].plot(history.history['accuracy'])
ax[0].plot(history.history['val_accuracy'])
ax[0].set_title('Training Accuracy vs Validation Accuracy')
ax[0].set_ylabel('Accuracy')
ax[0].set_xlabel('Epoch')
ax[0].legend(['Train', 'Validation'], loc='upper left')
ax[1].plot(history.history['loss'])
ax[1].plot(history.history['val_loss'])
ax[1].set_title('Training Loss vs Validation Loss')
ax[1].set_ylabel('Loss')
ax[1].set_xlabel('Epoch')
ax[1].legend(['Train', 'Validation'], loc='upper left')
plt.show()
fig1 = plt.gcf()
plt.plot(history.history['accuracy'])
plt.axis(ymin=0.4,ymax=1)
plt.grid()
plt.title('Model Accuracy')
plt.ylabel('Accuracy')
plt.xlabel('Epochs')
plt.legend(['train'])
plt.show()
model1.save('model_1.h5')
model1.save_weights('model_weights.h5')
from tensorflow.keras.optimizers import Adam
from tensorflow.keras.layers import Conv2D, MaxPool2D, Dropout, BatchNormalization, Flatten, Dense
from tensorflow.keras import regularizers
from tensorflow.keras.preprocessing.image import ImageDataGenerator
model_2 = tf.keras.models.Sequential([
Conv2D(32, kernel_size=(3, 3), padding='same', activation='relu', input_shape=(48, 48, 1)),
Conv2D(64, (3,3), padding='same', activation='relu'),
BatchNormalization(),
MaxPool2D(pool_size=(2, 2)),
Dropout(0.3),
Conv2D(128, (5,5), padding='same', activation='relu'),
BatchNormalization(),
MaxPool2D(pool_size=(2, 2)),
Dropout(0.4),
Conv2D(256, (3,3), padding='same', activation='relu', kernel_regularizer=regularizers.l2(0.01)),
BatchNormalization(),
MaxPool2D(pool_size=(2, 2)),
Dropout(0.5),
Flatten(),
Dense(256, activation='relu'),
BatchNormalization(),
Dropout(0.5),
Dense(128, activation='relu'),
BatchNormalization(),
Dropout(0.5),
Dense(7, activation='softmax')
])
optimizer = Adam(lr=0.001)
model_2.compile(optimizer=optimizer, loss='categorical_crossentropy', metrics=['accuracy'])
WARNING:absl:`lr` is deprecated, please use `learning_rate` instead, or use the legacy optimizer, e.g.,tf.keras.optimizers.legacy.Adam.
train_datagen = ImageDataGenerator(
rescale=1./255,
rotation_range=10,
zoom_range=0.1,
horizontal_flip=True
)
validation_datagen = ImageDataGenerator(
rescale=1./255
)
train_generator = train_datagen.flow_from_directory(
'/content/archive/train',
target_size=(48, 48),
batch_size=64,
color_mode='grayscale',
class_mode='categorical'
)
validation_generator = validation_datagen.flow_from_directory(
'/content/archive/test',
target_size=(48, 48),
batch_size=64,
color_mode='grayscale',
class_mode='categorical'
)
history_2 = model_2.fit(
train_generator,
steps_per_epoch=train_generator.samples/train_generator.batch_size,
epochs=5,
validation_data=validation_generator,
validation_steps=validation_generator.samples/validation_generator.batch_size,
verbose=1
)
Found 28709 images belonging to 7 classes. Found 7178 images belonging to 7 classes. Epoch 1/5 448/448 [==============================] - 37s 83ms/step - loss: 2.2919 - accuracy: 0.2979 - val_loss: 2.3117 - val_accuracy: 0.3217 Epoch 2/5 448/448 [==============================] - 37s 82ms/step - loss: 1.7883 - accuracy: 0.3959 - val_loss: 1.7279 - val_accuracy: 0.4071 Epoch 3/5 448/448 [==============================] - 35s 79ms/step - loss: 1.7023 - accuracy: 0.4387 - val_loss: 1.7077 - val_accuracy: 0.4280 Epoch 4/5 448/448 [==============================] - 36s 80ms/step - loss: 1.6777 - accuracy: 0.4614 - val_loss: 1.6121 - val_accuracy: 0.4882 Epoch 5/5 448/448 [==============================] - 36s 79ms/step - loss: 1.6506 - accuracy: 0.4749 - val_loss: 1.5657 - val_accuracy: 0.5104
fig1 = plt.gcf()
plt.plot(history_2.history['accuracy'])
plt.axis(ymin=0.4,ymax=1)
plt.grid()
plt.title('Model Accuracy')
plt.ylabel('Accuracy')
plt.xlabel('Epochs')
plt.legend(['train'])
plt.show()
fig , ax = plt.subplots(1,2)
train_acc = history_2.history['accuracy']
train_loss = history_2.history['loss']
fig.set_size_inches(12,4)
ax[0].plot(history_2.history['accuracy'])
ax[0].plot(history_2.history['val_accuracy'])
ax[0].set_title('Training Accuracy vs Validation Accuracy')
ax[0].set_ylabel('Accuracy')
ax[0].set_xlabel('Epoch')
ax[0].legend(['Train', 'Validation'], loc='upper left')
ax[1].plot(history_2.history['loss'])
ax[1].plot(history_2.history['val_loss'])
ax[1].set_title('Training Loss vs Validation Loss')
ax[1].set_ylabel('Loss')
ax[1].set_xlabel('Epoch')
ax[1].legend(['Train', 'Validation'], loc='upper left')
plt.show()
It is clear from the Accuracy and Loss Curve that I am not getting accuracy I am looking for.
# Evaluate the model on the test data
test_loss, test_acc = model_2.evaluate(train_generator)
print('Test accuracy:', test_acc)
449/449 [==============================] - 22s 48ms/step - loss: 1.4520 - accuracy: 0.6213 Test accuracy: 0.6212685704231262
# Evaluate the model on the test data
test_loss, test_acc = model_2.evaluate(validation_generator)
print('Test accuracy:', test_acc)
113/113 [==============================] - 3s 29ms/step - loss: 1.5065 - accuracy: 0.6039 Test accuracy: 0.6039286851882935
model_3 = tf.keras.models.Sequential([
Conv2D(32, kernel_size=(3, 3), padding='same', activation='swish', input_shape=(48, 48, 1)),
Conv2D(64, (3,3), padding='same', activation='relu'),
BatchNormalization(),
MaxPool2D(pool_size=(2, 2)),
Dropout(0.3),
Conv2D(128, (5,5), padding='same', activation='swish'),
BatchNormalization(),
MaxPool2D(pool_size=(2, 2)),
Dropout(0.4),
Conv2D(512, (3,3), padding='same', activation='swish', kernel_regularizer=regularizers.l2(0.01)),
BatchNormalization(),
MaxPool2D(pool_size=(2, 2)),
Dropout(0.5),
Flatten(),
Dense(256, activation='swish'),
BatchNormalization(),
Dropout(0.5),
Dense(128, activation='swish'),
BatchNormalization(),
Dropout(0.5),
Dense(7, activation='softmax')
])
model_3.summary()
Model: "sequential_8"
_________________________________________________________________
Layer (type) Output Shape Param #
=================================================================
conv2d_28 (Conv2D) (None, 48, 48, 32) 320
conv2d_29 (Conv2D) (None, 48, 48, 64) 18496
batch_normalization_29 (Bat (None, 48, 48, 64) 256
chNormalization)
max_pooling2d_20 (MaxPoolin (None, 24, 24, 64) 0
g2D)
dropout_29 (Dropout) (None, 24, 24, 64) 0
conv2d_30 (Conv2D) (None, 24, 24, 128) 204928
batch_normalization_30 (Bat (None, 24, 24, 128) 512
chNormalization)
max_pooling2d_21 (MaxPoolin (None, 12, 12, 128) 0
g2D)
dropout_30 (Dropout) (None, 12, 12, 128) 0
conv2d_31 (Conv2D) (None, 12, 12, 512) 590336
batch_normalization_31 (Bat (None, 12, 12, 512) 2048
chNormalization)
max_pooling2d_22 (MaxPoolin (None, 6, 6, 512) 0
g2D)
dropout_31 (Dropout) (None, 6, 6, 512) 0
flatten_7 (Flatten) (None, 18432) 0
dense_20 (Dense) (None, 256) 4718848
batch_normalization_32 (Bat (None, 256) 1024
chNormalization)
dropout_32 (Dropout) (None, 256) 0
dense_21 (Dense) (None, 128) 32896
batch_normalization_33 (Bat (None, 128) 512
chNormalization)
dropout_33 (Dropout) (None, 128) 0
dense_22 (Dense) (None, 7) 903
=================================================================
Total params: 5,571,079
Trainable params: 5,568,903
Non-trainable params: 2,176
_________________________________________________________________
optimizer = Adam(lr=0.001)
model_3.compile(optimizer=optimizer, loss='categorical_crossentropy', metrics=['accuracy'])
WARNING:absl:`lr` is deprecated, please use `learning_rate` instead, or use the legacy optimizer, e.g.,tf.keras.optimizers.legacy.Adam.
epochs= 60
batch_size=50
history_3 = model_3.fit(x = train_generator,epochs = epochs,validation_data = validation_generator)
Epoch 1/60 449/449 [==============================] - 31s 69ms/step - loss: 2.0706 - accuracy: 0.3176 - val_loss: 1.9007 - val_accuracy: 0.3129 Epoch 2/60 449/449 [==============================] - 30s 67ms/step - loss: 1.8046 - accuracy: 0.4144 - val_loss: 1.7169 - val_accuracy: 0.4666 Epoch 3/60 449/449 [==============================] - 30s 67ms/step - loss: 1.7604 - accuracy: 0.4599 - val_loss: 1.6564 - val_accuracy: 0.5014 Epoch 4/60 449/449 [==============================] - 30s 67ms/step - loss: 1.7344 - accuracy: 0.4794 - val_loss: 1.6645 - val_accuracy: 0.5127 Epoch 5/60 449/449 [==============================] - 29s 64ms/step - loss: 1.7379 - accuracy: 0.4930 - val_loss: 1.6922 - val_accuracy: 0.5230 Epoch 6/60 449/449 [==============================] - 30s 67ms/step - loss: 1.7140 - accuracy: 0.4994 - val_loss: 1.6086 - val_accuracy: 0.5371 Epoch 7/60 449/449 [==============================] - 68s 151ms/step - loss: 1.6905 - accuracy: 0.5120 - val_loss: 1.6370 - val_accuracy: 0.5446 Epoch 8/60 449/449 [==============================] - 31s 68ms/step - loss: 1.7093 - accuracy: 0.5126 - val_loss: 1.6072 - val_accuracy: 0.5443 Epoch 9/60 449/449 [==============================] - 30s 66ms/step - loss: 1.6886 - accuracy: 0.5219 - val_loss: 1.5672 - val_accuracy: 0.5497 Epoch 10/60 449/449 [==============================] - 29s 64ms/step - loss: 1.6747 - accuracy: 0.5257 - val_loss: 1.5592 - val_accuracy: 0.5556 Epoch 11/60 449/449 [==============================] - 33s 73ms/step - loss: 1.6697 - accuracy: 0.5303 - val_loss: 1.6010 - val_accuracy: 0.5553 Epoch 12/60 449/449 [==============================] - 29s 65ms/step - loss: 1.6636 - accuracy: 0.5350 - val_loss: 1.5484 - val_accuracy: 0.5676 Epoch 13/60 449/449 [==============================] - 32s 72ms/step - loss: 1.6989 - accuracy: 0.5343 - val_loss: 1.6207 - val_accuracy: 0.5680 Epoch 14/60 449/449 [==============================] - 29s 64ms/step - loss: 1.6518 - accuracy: 0.5336 - val_loss: 1.5602 - val_accuracy: 0.5645 Epoch 15/60 449/449 [==============================] - 30s 66ms/step - loss: 1.6394 - accuracy: 0.5394 - val_loss: 1.5505 - val_accuracy: 0.5645 Epoch 16/60 449/449 [==============================] - 31s 68ms/step - loss: 1.6476 - accuracy: 0.5398 - val_loss: 1.6532 - val_accuracy: 0.5454 Epoch 17/60 449/449 [==============================] - 30s 66ms/step - loss: 1.6340 - accuracy: 0.5469 - val_loss: 1.5407 - val_accuracy: 0.5702 Epoch 18/60 449/449 [==============================] - 29s 64ms/step - loss: 1.6505 - accuracy: 0.5475 - val_loss: 1.5669 - val_accuracy: 0.5811 Epoch 19/60 449/449 [==============================] - 31s 68ms/step - loss: 1.7408 - accuracy: 0.5437 - val_loss: 1.7698 - val_accuracy: 0.5617 Epoch 20/60 449/449 [==============================] - 29s 65ms/step - loss: 1.7147 - accuracy: 0.5476 - val_loss: 1.5556 - val_accuracy: 0.5780 Epoch 21/60 449/449 [==============================] - 30s 67ms/step - loss: 1.6709 - accuracy: 0.5499 - val_loss: 1.5499 - val_accuracy: 0.5765 Epoch 22/60 449/449 [==============================] - 29s 65ms/step - loss: 1.6733 - accuracy: 0.5534 - val_loss: 1.5583 - val_accuracy: 0.5841 Epoch 23/60 449/449 [==============================] - 29s 65ms/step - loss: 1.6630 - accuracy: 0.5541 - val_loss: 1.5890 - val_accuracy: 0.5756 Epoch 24/60 449/449 [==============================] - 30s 66ms/step - loss: 1.6690 - accuracy: 0.5566 - val_loss: 1.6052 - val_accuracy: 0.5626 Epoch 25/60 449/449 [==============================] - 30s 67ms/step - loss: 1.6221 - accuracy: 0.5608 - val_loss: 1.5023 - val_accuracy: 0.5807 Epoch 26/60 449/449 [==============================] - 31s 70ms/step - loss: 1.6789 - accuracy: 0.5610 - val_loss: 1.6842 - val_accuracy: 0.5770 Epoch 27/60 449/449 [==============================] - 29s 65ms/step - loss: 1.6757 - accuracy: 0.5613 - val_loss: 1.6325 - val_accuracy: 0.5649 Epoch 28/60 449/449 [==============================] - 30s 67ms/step - loss: 1.5996 - accuracy: 0.5633 - val_loss: 1.5822 - val_accuracy: 0.5656 Epoch 29/60 449/449 [==============================] - 29s 65ms/step - loss: 1.6102 - accuracy: 0.5656 - val_loss: 1.4973 - val_accuracy: 0.5887 Epoch 30/60 449/449 [==============================] - 31s 69ms/step - loss: 1.6253 - accuracy: 0.5671 - val_loss: 1.5022 - val_accuracy: 0.5935 Epoch 31/60 449/449 [==============================] - 29s 64ms/step - loss: 1.5886 - accuracy: 0.5708 - val_loss: 1.6365 - val_accuracy: 0.5670 Epoch 32/60 449/449 [==============================] - 31s 70ms/step - loss: 1.6985 - accuracy: 0.5613 - val_loss: 1.6364 - val_accuracy: 0.5978 Epoch 33/60 449/449 [==============================] - 29s 64ms/step - loss: 1.6741 - accuracy: 0.5651 - val_loss: 1.5484 - val_accuracy: 0.5913 Epoch 34/60 449/449 [==============================] - 30s 66ms/step - loss: 1.9269 - accuracy: 0.5583 - val_loss: 1.9458 - val_accuracy: 0.5637 Epoch 35/60 449/449 [==============================] - 32s 71ms/step - loss: 1.7627 - accuracy: 0.5643 - val_loss: 1.5818 - val_accuracy: 0.5925 Epoch 36/60 449/449 [==============================] - 30s 67ms/step - loss: 1.6701 - accuracy: 0.5699 - val_loss: 1.8428 - val_accuracy: 0.5854 Epoch 37/60 449/449 [==============================] - 29s 66ms/step - loss: 1.6115 - accuracy: 0.5780 - val_loss: 1.5065 - val_accuracy: 0.5872 Epoch 38/60 449/449 [==============================] - 30s 66ms/step - loss: 1.6093 - accuracy: 0.5784 - val_loss: 1.5191 - val_accuracy: 0.5929 Epoch 39/60 449/449 [==============================] - 30s 66ms/step - loss: 1.6455 - accuracy: 0.5746 - val_loss: 1.5397 - val_accuracy: 0.5977 Epoch 40/60 449/449 [==============================] - 30s 67ms/step - loss: 1.6010 - accuracy: 0.5761 - val_loss: 1.4837 - val_accuracy: 0.5972 Epoch 41/60 449/449 [==============================] - 30s 67ms/step - loss: 1.5637 - accuracy: 0.5778 - val_loss: 1.4899 - val_accuracy: 0.5982 Epoch 42/60 449/449 [==============================] - 29s 65ms/step - loss: 1.5735 - accuracy: 0.5749 - val_loss: 1.4818 - val_accuracy: 0.5968 Epoch 43/60 449/449 [==============================] - 30s 67ms/step - loss: 1.5892 - accuracy: 0.5793 - val_loss: 1.5609 - val_accuracy: 0.6031 Epoch 44/60 449/449 [==============================] - 30s 67ms/step - loss: 1.6049 - accuracy: 0.5794 - val_loss: 1.5816 - val_accuracy: 0.5756 Epoch 45/60 449/449 [==============================] - 29s 64ms/step - loss: 1.6448 - accuracy: 0.5763 - val_loss: 1.5766 - val_accuracy: 0.5908 Epoch 46/60 449/449 [==============================] - 30s 66ms/step - loss: 1.5680 - accuracy: 0.5793 - val_loss: 1.4561 - val_accuracy: 0.6017 Epoch 47/60 449/449 [==============================] - 31s 68ms/step - loss: 1.6357 - accuracy: 0.5794 - val_loss: 1.5130 - val_accuracy: 0.5938 Epoch 48/60 449/449 [==============================] - 30s 66ms/step - loss: 1.5624 - accuracy: 0.5872 - val_loss: 1.5598 - val_accuracy: 0.6048 Epoch 49/60 449/449 [==============================] - 29s 65ms/step - loss: 1.6062 - accuracy: 0.5816 - val_loss: 1.6514 - val_accuracy: 0.6028 Epoch 50/60 449/449 [==============================] - 31s 68ms/step - loss: 1.6088 - accuracy: 0.5845 - val_loss: 1.5405 - val_accuracy: 0.6021 Epoch 51/60 449/449 [==============================] - 29s 64ms/step - loss: 1.5924 - accuracy: 0.5809 - val_loss: 1.5389 - val_accuracy: 0.5963 Epoch 52/60 449/449 [==============================] - 30s 66ms/step - loss: 1.5493 - accuracy: 0.5858 - val_loss: 1.5425 - val_accuracy: 0.5758 Epoch 53/60 449/449 [==============================] - 32s 71ms/step - loss: 1.5712 - accuracy: 0.5859 - val_loss: 1.4967 - val_accuracy: 0.5993 Epoch 54/60 449/449 [==============================] - 29s 64ms/step - loss: 1.5721 - accuracy: 0.5874 - val_loss: 1.4550 - val_accuracy: 0.6027 Epoch 55/60 449/449 [==============================] - 32s 70ms/step - loss: 1.5841 - accuracy: 0.5873 - val_loss: 1.6010 - val_accuracy: 0.6119 Epoch 56/60 449/449 [==============================] - 29s 64ms/step - loss: 1.6508 - accuracy: 0.5844 - val_loss: 1.5471 - val_accuracy: 0.6046 Epoch 57/60 449/449 [==============================] - 30s 67ms/step - loss: 1.5733 - accuracy: 0.5887 - val_loss: 1.5251 - val_accuracy: 0.5890 Epoch 58/60 449/449 [==============================] - 32s 71ms/step - loss: 1.6055 - accuracy: 0.5914 - val_loss: 1.4596 - val_accuracy: 0.5986 Epoch 59/60 449/449 [==============================] - 30s 67ms/step - loss: 1.5428 - accuracy: 0.5908 - val_loss: 1.4837 - val_accuracy: 0.6031 Epoch 60/60 449/449 [==============================] - 33s 73ms/step - loss: 1.5783 - accuracy: 0.5902 - val_loss: 1.5603 - val_accuracy: 0.5995
fig2 = plt.gcf()
plt.plot(history_3.history['accuracy'])
plt.axis(ymin=0.4,ymax=1)
plt.grid()
plt.title('Model Accuracy')
plt.ylabel('Accuracy')
plt.xlabel('Epochs')
plt.legend(['train'])
plt.show()
fig2 , ax = plt.subplots(1,2)
train_acc = history_3.history['accuracy']
train_loss = history_3.history['loss']
fig2.set_size_inches(12,4)
ax[0].plot(history_3.history['accuracy'])
ax[0].plot(history_3.history['val_accuracy'])
ax[0].set_title('Training Accuracy vs Validation Accuracy')
ax[0].set_ylabel('Accuracy')
ax[0].set_xlabel('Epoch')
ax[0].legend(['Train', 'Validation'], loc='upper left')
ax[1].plot(history_3.history['loss'])
ax[1].plot(history_3.history['val_loss'])
ax[1].set_title('Training Loss vs Validation Loss')
ax[1].set_ylabel('Loss')
ax[1].set_xlabel('Epoch')
ax[1].legend(['Train', 'Validation'], loc='upper left')
plt.show()
# Evaluate the model on the test data
test_loss, test_acc = model_3.evaluate(train_generator)
print('Test accuracy:', test_acc)
449/449 [==============================] - 22s 50ms/step - loss: 1.4892 - accuracy: 0.6271 Test accuracy: 0.6271204352378845
# Evaluate the model on the test data
test_loss, test_acc = model_3.evaluate(validation_generator)
print('Test accuracy:', test_acc)
113/113 [==============================] - 3s 23ms/step - loss: 1.5603 - accuracy: 0.5995 Test accuracy: 0.5994706153869629
# Extracting the dataset
zip_ref = zipfile.ZipFile("archive.zip", "r")
zip_ref.extractall()
zip_ref.close()
dataset_path = "/content/archive"
labels = os.listdir(dataset_path)
print(labels)
for label in labels:
print(os.listdir(os.path.join(dataset_path, label)))
# Preparing training and validation directories
train_dir = "/content/archive/train"
test_dir = "/content/archive/test"
# Defining ImageDataGenerator for data augmentation
img_size = 48
train_datagen = ImageDataGenerator(
rescale = 1./255,
rotation_range = 20,
zoom_range = 0.2,
width_shift_range = 0.1,
height_shift_range = 0.1,
shear_range = 0.2,
horizontal_flip = True,
vertical_flip = True,
fill_mode = 'nearest',
validation_split = 0.2
)
# Preparing training data using ImageDataGenerator
train_generator = train_datagen.flow_from_directory(
directory = train_dir,
target_size = (img_size, img_size),
batch_size = 64,
color_mode = "grayscale",
class_mode = "categorical",
subset = "training"
)
# Preparing validation data using ImageDataGenerator
validation_datagen = ImageDataGenerator(
rescale = 1./255,
validation_split = 0.2
)
validation_generator = validation_datagen.flow_from_directory(
directory = test_dir,
target_size = (img_size, img_size),
batch_size = 64,
color_mode = "grayscale",
class_mode = "categorical",
subset = "validation"
)
# Loading an image for visualization
img = Image.open('/content/archive/test/angry/im106.png')
img.show()
img = load_img('/content/archive/test/angry/im106.png', target_size=(48, 48), color_mode="grayscale")
img = np.array(img)
plt.imshow(img)
print(img.shape)
# Preparing training data using tf.keras.preprocessing.image_dataset_from_directory
train_ds = tf.keras.preprocessing.image_dataset_from_directory(
directory = dataset_path,
validation_split = 0.2,
subset = "training",
seed = 123,
image_size = (224, 224),
batch_size = 32
)
['test', 'train'] ['fearful', 'disgusted', 'angry', 'neutral', 'surprised', 'sad', 'happy'] ['fearful', 'disgusted', 'angry', 'neutral', 'surprised', 'sad', 'happy'] Found 22968 images belonging to 7 classes. Found 1432 images belonging to 7 classes.
(48, 48) Found 35887 files belonging to 2 classes. Using 28710 files for training.
from tensorflow.keras.preprocessing.image import load_img, img_to_array
img = load_img("../content/archive/test/fearful/im1002.png",target_size = (48,48),color_mode = "grayscale")
img = np.array(img)
plt.imshow(img)
print(img.shape)
import cv2
import numpy as np
label_dict = {0:'Angry',1:'Disgust',2:'Fear',3:'Happy',4:'Neutral',5:'Sad',6:'Surprise'}
img = cv2.imread('/content/archive/test/fearful/im1002.png', cv2.IMREAD_GRAYSCALE)
if img is None:
raise ValueError('Failed to load image. Check if image path is correct and image file exists.')
elif img.size == 0:
raise ValueError('Input image is empty or has invalid dimensions. Check the image file.')
img = cv2.resize(img, (227, 227))
img = np.repeat(img[..., np.newaxis], 3, -1)
result = model.predict(np.array([img]))
result = list(result[0])
print(result)
img_index = result.index(max(result))
print(label_dict[img_index])
plt.imshow(img)
plt.show()
(48, 48) 1/1 [==============================] - 0s 22ms/step [0.0, 0.0, 9.047387e-16, 0.0, 1.0, 0.0, 0.0] Neutral
# Creating and compiling the model
from tensorflow.keras.optimizers import Adam, RMSprop, SGD, Adamax
model_4 = Sequential()
model_4.add(Conv2D(32, kernel_size = (3, 3), padding = 'same', activation = 'relu', input_shape = (48, 48, 1)))
model_4.add(Conv2D(64, (3, 3), padding = 'same', activation = 'relu'))
model_4.add(BatchNormalization())
model_4.add(MaxPool2D(pool_size = (2, 2)))
model_4.add(Dropout(0.25))
model_4.add(Conv2D(128, (5, 5), padding = 'same', activation = 'relu'))
model_4.add(Conv2D(256, (5, 5), padding = 'same', activation = 'relu'))
model_4.add(Conv2D(512, (5, 5), padding = 'same', activation = 'swish'))
model_4.add(BatchNormalization())
model_4.add(MaxPool2D(pool_size = (2, 2)))
model_4.add(Dropout(0.25))
model_4.add(Flatten()),
model_4.add(Dense(256, activation='relu'))
model_4.add(BatchNormalization())
model_4.add(Dropout(0.5))
model_4.add(Dense(128, activation = 'relu'))
model_4.add(BatchNormalization())
model_4.add(Dropout(0.5))
model_4.add(Dense(7, activation = 'softmax'))
model_4.summary()
Model: "sequential_15"
_________________________________________________________________
Layer (type) Output Shape Param #
=================================================================
conv2d_60 (Conv2D) (None, 48, 48, 32) 320
conv2d_61 (Conv2D) (None, 48, 48, 64) 18496
batch_normalization_56 (Bat (None, 48, 48, 64) 256
chNormalization)
max_pooling2d_35 (MaxPoolin (None, 24, 24, 64) 0
g2D)
dropout_56 (Dropout) (None, 24, 24, 64) 0
conv2d_62 (Conv2D) (None, 24, 24, 128) 204928
conv2d_63 (Conv2D) (None, 24, 24, 256) 819456
conv2d_64 (Conv2D) (None, 24, 24, 512) 3277312
batch_normalization_57 (Bat (None, 24, 24, 512) 2048
chNormalization)
max_pooling2d_36 (MaxPoolin (None, 12, 12, 512) 0
g2D)
dropout_57 (Dropout) (None, 12, 12, 512) 0
flatten_14 (Flatten) (None, 73728) 0
dense_39 (Dense) (None, 256) 18874624
batch_normalization_58 (Bat (None, 256) 1024
chNormalization)
dropout_58 (Dropout) (None, 256) 0
dense_40 (Dense) (None, 128) 32896
batch_normalization_59 (Bat (None, 128) 512
chNormalization)
dropout_59 (Dropout) (None, 128) 0
dense_41 (Dense) (None, 7) 903
=================================================================
Total params: 23,232,775
Trainable params: 23,230,855
Non-trainable params: 1,920
_________________________________________________________________
optimizer = Adam(lr=0.001)
model_4.compile(optimizer = optimizer, loss = 'categorical_crossentropy', metrics = ['accuracy'])
WARNING:absl:`lr` is deprecated, please use `learning_rate` instead, or use the legacy optimizer, e.g.,tf.keras.optimizers.legacy.Adam.
epochs= 100
batch_size=40
history_4 = model_4.fit(x = train_generator,epochs = epochs,validation_data = validation_generator)
Epoch 1/100 359/359 [==============================] - 77s 94ms/step - loss: 2.2728 - accuracy: 0.1957 - val_loss: 1.8245 - val_accuracy: 0.2465 Epoch 2/100 359/359 [==============================] - 34s 96ms/step - loss: 1.8999 - accuracy: 0.2323 - val_loss: 1.7758 - val_accuracy: 0.2689 Epoch 3/100 359/359 [==============================] - 34s 95ms/step - loss: 1.8164 - accuracy: 0.2522 - val_loss: 1.7802 - val_accuracy: 0.2709 Epoch 4/100 359/359 [==============================] - 33s 92ms/step - loss: 1.7824 - accuracy: 0.2668 - val_loss: 1.9309 - val_accuracy: 0.2493 Epoch 5/100 359/359 [==============================] - 35s 97ms/step - loss: 1.7334 - accuracy: 0.2985 - val_loss: 1.6452 - val_accuracy: 0.3331 Epoch 6/100 359/359 [==============================] - 34s 95ms/step - loss: 1.6903 - accuracy: 0.3207 - val_loss: 1.6060 - val_accuracy: 0.3624 Epoch 7/100 359/359 [==============================] - 34s 95ms/step - loss: 1.6508 - accuracy: 0.3440 - val_loss: 1.6774 - val_accuracy: 0.3457 Epoch 8/100 359/359 [==============================] - 34s 93ms/step - loss: 1.6055 - accuracy: 0.3670 - val_loss: 1.6336 - val_accuracy: 0.3904 Epoch 9/100 359/359 [==============================] - 33s 93ms/step - loss: 1.5296 - accuracy: 0.4047 - val_loss: 1.5944 - val_accuracy: 0.3659 Epoch 10/100 359/359 [==============================] - 33s 93ms/step - loss: 1.4910 - accuracy: 0.4252 - val_loss: 1.4015 - val_accuracy: 0.4721 Epoch 11/100 359/359 [==============================] - 33s 93ms/step - loss: 1.4289 - accuracy: 0.4460 - val_loss: 1.4617 - val_accuracy: 0.4525 Epoch 12/100 359/359 [==============================] - 33s 93ms/step - loss: 1.3832 - accuracy: 0.4688 - val_loss: 1.3259 - val_accuracy: 0.4916 Epoch 13/100 359/359 [==============================] - 33s 93ms/step - loss: 1.3484 - accuracy: 0.4868 - val_loss: 1.2273 - val_accuracy: 0.5363 Epoch 14/100 359/359 [==============================] - 33s 92ms/step - loss: 1.3200 - accuracy: 0.4969 - val_loss: 1.2162 - val_accuracy: 0.5300 Epoch 15/100 359/359 [==============================] - 33s 93ms/step - loss: 1.2920 - accuracy: 0.5057 - val_loss: 1.2331 - val_accuracy: 0.5328 Epoch 16/100 359/359 [==============================] - 33s 93ms/step - loss: 1.2611 - accuracy: 0.5206 - val_loss: 1.1751 - val_accuracy: 0.5461 Epoch 17/100 359/359 [==============================] - 33s 93ms/step - loss: 1.2468 - accuracy: 0.5269 - val_loss: 1.2542 - val_accuracy: 0.5244 Epoch 18/100 359/359 [==============================] - 33s 92ms/step - loss: 1.2401 - accuracy: 0.5282 - val_loss: 1.1372 - val_accuracy: 0.5817 Epoch 19/100 359/359 [==============================] - 33s 93ms/step - loss: 1.2172 - accuracy: 0.5351 - val_loss: 1.1293 - val_accuracy: 0.5594 Epoch 20/100 359/359 [==============================] - 33s 91ms/step - loss: 1.1996 - accuracy: 0.5475 - val_loss: 1.1713 - val_accuracy: 0.5482 Epoch 21/100 359/359 [==============================] - 33s 93ms/step - loss: 1.1905 - accuracy: 0.5515 - val_loss: 1.0890 - val_accuracy: 0.5908 Epoch 22/100 359/359 [==============================] - 33s 93ms/step - loss: 1.1770 - accuracy: 0.5549 - val_loss: 1.1273 - val_accuracy: 0.5740 Epoch 23/100 359/359 [==============================] - 33s 93ms/step - loss: 1.1650 - accuracy: 0.5599 - val_loss: 1.0763 - val_accuracy: 0.5999 Epoch 24/100 359/359 [==============================] - 33s 93ms/step - loss: 1.1462 - accuracy: 0.5687 - val_loss: 1.1067 - val_accuracy: 0.5698 Epoch 25/100 359/359 [==============================] - 33s 93ms/step - loss: 1.1454 - accuracy: 0.5691 - val_loss: 1.0720 - val_accuracy: 0.6138 Epoch 26/100 359/359 [==============================] - 33s 93ms/step - loss: 1.1379 - accuracy: 0.5730 - val_loss: 1.0705 - val_accuracy: 0.6027 Epoch 27/100 359/359 [==============================] - 33s 92ms/step - loss: 1.1240 - accuracy: 0.5764 - val_loss: 1.0827 - val_accuracy: 0.5950 Epoch 28/100 359/359 [==============================] - 33s 92ms/step - loss: 1.1210 - accuracy: 0.5775 - val_loss: 1.1011 - val_accuracy: 0.5999 Epoch 29/100 359/359 [==============================] - 34s 95ms/step - loss: 1.1022 - accuracy: 0.5853 - val_loss: 1.1103 - val_accuracy: 0.5957 Epoch 30/100 359/359 [==============================] - 33s 91ms/step - loss: 1.1021 - accuracy: 0.5845 - val_loss: 1.1183 - val_accuracy: 0.5719 Epoch 31/100 359/359 [==============================] - 33s 93ms/step - loss: 1.0856 - accuracy: 0.5921 - val_loss: 1.0873 - val_accuracy: 0.5873 Epoch 32/100 359/359 [==============================] - 33s 92ms/step - loss: 1.0877 - accuracy: 0.5904 - val_loss: 1.0712 - val_accuracy: 0.5859 Epoch 33/100 359/359 [==============================] - 33s 92ms/step - loss: 1.0835 - accuracy: 0.5949 - val_loss: 1.0829 - val_accuracy: 0.5859 Epoch 34/100 359/359 [==============================] - 34s 93ms/step - loss: 1.0744 - accuracy: 0.5981 - val_loss: 1.0483 - val_accuracy: 0.6145 Epoch 35/100 359/359 [==============================] - 34s 94ms/step - loss: 1.0687 - accuracy: 0.6016 - val_loss: 1.0257 - val_accuracy: 0.6264 Epoch 36/100 359/359 [==============================] - 33s 93ms/step - loss: 1.0642 - accuracy: 0.6004 - val_loss: 1.1158 - val_accuracy: 0.5803 Epoch 37/100 359/359 [==============================] - 33s 90ms/step - loss: 1.0520 - accuracy: 0.6088 - val_loss: 1.0566 - val_accuracy: 0.6089 Epoch 38/100 359/359 [==============================] - 33s 92ms/step - loss: 1.0487 - accuracy: 0.6111 - val_loss: 1.0712 - val_accuracy: 0.6020 Epoch 39/100 359/359 [==============================] - 33s 92ms/step - loss: 1.0463 - accuracy: 0.6071 - val_loss: 1.0440 - val_accuracy: 0.6152 Epoch 40/100 359/359 [==============================] - 33s 91ms/step - loss: 1.0410 - accuracy: 0.6087 - val_loss: 1.0140 - val_accuracy: 0.6292 Epoch 41/100 359/359 [==============================] - 33s 93ms/step - loss: 1.0245 - accuracy: 0.6179 - val_loss: 1.0499 - val_accuracy: 0.6096 Epoch 42/100 359/359 [==============================] - 33s 92ms/step - loss: 1.0228 - accuracy: 0.6156 - val_loss: 0.9943 - val_accuracy: 0.6229 Epoch 43/100 359/359 [==============================] - 33s 92ms/step - loss: 1.0216 - accuracy: 0.6193 - val_loss: 1.0383 - val_accuracy: 0.6117 Epoch 44/100 359/359 [==============================] - 33s 92ms/step - loss: 1.0173 - accuracy: 0.6203 - val_loss: 1.0328 - val_accuracy: 0.6173 Epoch 45/100 359/359 [==============================] - 33s 93ms/step - loss: 1.0163 - accuracy: 0.6237 - val_loss: 1.0508 - val_accuracy: 0.5964 Epoch 46/100 359/359 [==============================] - 33s 91ms/step - loss: 1.0081 - accuracy: 0.6235 - val_loss: 1.0523 - val_accuracy: 0.6041 Epoch 47/100 359/359 [==============================] - 34s 94ms/step - loss: 0.9984 - accuracy: 0.6279 - val_loss: 0.9930 - val_accuracy: 0.6243 Epoch 48/100 359/359 [==============================] - 33s 93ms/step - loss: 0.9895 - accuracy: 0.6320 - val_loss: 0.9713 - val_accuracy: 0.6299 Epoch 49/100 359/359 [==============================] - 33s 93ms/step - loss: 0.9965 - accuracy: 0.6311 - val_loss: 1.0213 - val_accuracy: 0.6229 Epoch 50/100 359/359 [==============================] - 33s 91ms/step - loss: 0.9887 - accuracy: 0.6296 - val_loss: 0.9923 - val_accuracy: 0.6327 Epoch 51/100 359/359 [==============================] - 33s 92ms/step - loss: 0.9910 - accuracy: 0.6330 - val_loss: 1.0475 - val_accuracy: 0.5992 Epoch 52/100 359/359 [==============================] - 33s 92ms/step - loss: 0.9788 - accuracy: 0.6352 - val_loss: 0.9752 - val_accuracy: 0.6404 Epoch 53/100 359/359 [==============================] - 33s 92ms/step - loss: 0.9831 - accuracy: 0.6357 - val_loss: 0.9781 - val_accuracy: 0.6369 Epoch 54/100 359/359 [==============================] - 34s 93ms/step - loss: 0.9719 - accuracy: 0.6380 - val_loss: 1.0128 - val_accuracy: 0.6376 Epoch 55/100 359/359 [==============================] - 34s 93ms/step - loss: 0.9723 - accuracy: 0.6366 - val_loss: 0.9688 - val_accuracy: 0.6313 Epoch 56/100 359/359 [==============================] - 33s 93ms/step - loss: 0.9619 - accuracy: 0.6408 - val_loss: 1.0088 - val_accuracy: 0.6299 Epoch 57/100 359/359 [==============================] - 32s 90ms/step - loss: 0.9641 - accuracy: 0.6425 - val_loss: 0.9681 - val_accuracy: 0.6515 Epoch 58/100 359/359 [==============================] - 33s 93ms/step - loss: 0.9560 - accuracy: 0.6458 - val_loss: 0.9988 - val_accuracy: 0.6299 Epoch 59/100 359/359 [==============================] - 34s 94ms/step - loss: 0.9515 - accuracy: 0.6453 - val_loss: 0.9812 - val_accuracy: 0.6327 Epoch 60/100 359/359 [==============================] - 33s 92ms/step - loss: 0.9526 - accuracy: 0.6462 - val_loss: 1.0278 - val_accuracy: 0.6313 Epoch 61/100 359/359 [==============================] - 33s 93ms/step - loss: 0.9419 - accuracy: 0.6520 - val_loss: 0.9908 - val_accuracy: 0.6278 Epoch 62/100 359/359 [==============================] - 33s 93ms/step - loss: 0.9371 - accuracy: 0.6555 - val_loss: 0.9662 - val_accuracy: 0.6501 Epoch 63/100 359/359 [==============================] - 33s 93ms/step - loss: 0.9324 - accuracy: 0.6532 - val_loss: 1.0020 - val_accuracy: 0.6341 Epoch 64/100 359/359 [==============================] - 33s 93ms/step - loss: 0.9394 - accuracy: 0.6506 - val_loss: 0.9748 - val_accuracy: 0.6480 Epoch 65/100 359/359 [==============================] - 33s 93ms/step - loss: 0.9320 - accuracy: 0.6560 - val_loss: 0.9920 - val_accuracy: 0.6404 Epoch 66/100 359/359 [==============================] - 33s 92ms/step - loss: 0.9277 - accuracy: 0.6570 - val_loss: 0.9544 - val_accuracy: 0.6494 Epoch 67/100 359/359 [==============================] - 33s 93ms/step - loss: 0.9212 - accuracy: 0.6590 - val_loss: 0.9581 - val_accuracy: 0.6466 Epoch 68/100 359/359 [==============================] - 33s 91ms/step - loss: 0.9175 - accuracy: 0.6609 - val_loss: 0.9665 - val_accuracy: 0.6501 Epoch 69/100 359/359 [==============================] - 33s 92ms/step - loss: 0.9178 - accuracy: 0.6617 - val_loss: 0.9769 - val_accuracy: 0.6439 Epoch 70/100 359/359 [==============================] - 33s 92ms/step - loss: 0.9162 - accuracy: 0.6627 - val_loss: 1.0420 - val_accuracy: 0.6257 Epoch 71/100 359/359 [==============================] - 32s 90ms/step - loss: 0.9133 - accuracy: 0.6617 - val_loss: 1.0044 - val_accuracy: 0.6411 Epoch 72/100 359/359 [==============================] - 33s 92ms/step - loss: 0.9019 - accuracy: 0.6674 - val_loss: 0.9707 - val_accuracy: 0.6522 Epoch 73/100 359/359 [==============================] - 33s 92ms/step - loss: 0.9048 - accuracy: 0.6658 - val_loss: 0.9438 - val_accuracy: 0.6585 Epoch 74/100 359/359 [==============================] - 33s 91ms/step - loss: 0.9013 - accuracy: 0.6696 - val_loss: 0.9875 - val_accuracy: 0.6466 Epoch 75/100 359/359 [==============================] - 33s 91ms/step - loss: 0.9047 - accuracy: 0.6672 - val_loss: 0.9871 - val_accuracy: 0.6348 Epoch 76/100 359/359 [==============================] - 33s 92ms/step - loss: 0.8995 - accuracy: 0.6675 - val_loss: 0.9610 - val_accuracy: 0.6550 Epoch 77/100 359/359 [==============================] - 33s 91ms/step - loss: 0.8941 - accuracy: 0.6731 - val_loss: 0.9839 - val_accuracy: 0.6466 Epoch 78/100 359/359 [==============================] - 32s 90ms/step - loss: 0.8934 - accuracy: 0.6739 - val_loss: 1.0237 - val_accuracy: 0.6292 Epoch 79/100 359/359 [==============================] - 33s 92ms/step - loss: 0.8921 - accuracy: 0.6734 - val_loss: 0.9636 - val_accuracy: 0.6425 Epoch 80/100 359/359 [==============================] - 33s 92ms/step - loss: 0.8839 - accuracy: 0.6736 - val_loss: 0.9745 - val_accuracy: 0.6425 Epoch 81/100 359/359 [==============================] - 33s 91ms/step - loss: 0.8804 - accuracy: 0.6723 - val_loss: 0.9712 - val_accuracy: 0.6501 Epoch 82/100 359/359 [==============================] - 33s 91ms/step - loss: 0.8735 - accuracy: 0.6784 - val_loss: 0.9436 - val_accuracy: 0.6571 Epoch 83/100 359/359 [==============================] - 33s 93ms/step - loss: 0.8789 - accuracy: 0.6782 - val_loss: 1.0188 - val_accuracy: 0.6487 Epoch 84/100 359/359 [==============================] - 33s 91ms/step - loss: 0.8747 - accuracy: 0.6792 - val_loss: 1.0007 - val_accuracy: 0.6453 Epoch 85/100 359/359 [==============================] - 33s 92ms/step - loss: 0.8710 - accuracy: 0.6833 - val_loss: 0.9961 - val_accuracy: 0.6466 Epoch 86/100 359/359 [==============================] - 33s 91ms/step - loss: 0.8658 - accuracy: 0.6809 - val_loss: 0.9896 - val_accuracy: 0.6390 Epoch 87/100 359/359 [==============================] - 33s 92ms/step - loss: 0.8617 - accuracy: 0.6831 - val_loss: 1.0238 - val_accuracy: 0.6453 Epoch 88/100 359/359 [==============================] - 33s 92ms/step - loss: 0.8638 - accuracy: 0.6828 - val_loss: 0.9863 - val_accuracy: 0.6425 Epoch 89/100 359/359 [==============================] - 34s 94ms/step - loss: 0.8589 - accuracy: 0.6856 - val_loss: 0.9686 - val_accuracy: 0.6522 Epoch 90/100 359/359 [==============================] - 33s 91ms/step - loss: 0.8564 - accuracy: 0.6856 - val_loss: 0.9612 - val_accuracy: 0.6501 Epoch 91/100 359/359 [==============================] - 33s 91ms/step - loss: 0.8463 - accuracy: 0.6880 - val_loss: 0.9824 - val_accuracy: 0.6501 Epoch 92/100 359/359 [==============================] - 33s 92ms/step - loss: 0.8514 - accuracy: 0.6873 - val_loss: 1.0089 - val_accuracy: 0.6306 Epoch 93/100 359/359 [==============================] - 33s 92ms/step - loss: 0.8444 - accuracy: 0.6910 - val_loss: 0.9738 - val_accuracy: 0.6494 Epoch 94/100 359/359 [==============================] - 32s 90ms/step - loss: 0.8474 - accuracy: 0.6875 - val_loss: 0.9960 - val_accuracy: 0.6418 Epoch 95/100 359/359 [==============================] - 33s 93ms/step - loss: 0.8436 - accuracy: 0.6906 - val_loss: 0.9765 - val_accuracy: 0.6571 Epoch 96/100 359/359 [==============================] - 33s 92ms/step - loss: 0.8359 - accuracy: 0.6971 - val_loss: 0.9705 - val_accuracy: 0.6522 Epoch 97/100 359/359 [==============================] - 34s 94ms/step - loss: 0.8366 - accuracy: 0.6899 - val_loss: 0.9928 - val_accuracy: 0.6446 Epoch 98/100 359/359 [==============================] - 33s 93ms/step - loss: 0.8352 - accuracy: 0.6947 - val_loss: 0.9669 - val_accuracy: 0.6641 Epoch 99/100 359/359 [==============================] - 33s 93ms/step - loss: 0.8419 - accuracy: 0.6941 - val_loss: 1.0753 - val_accuracy: 0.6362 Epoch 100/100 359/359 [==============================] - 34s 94ms/step - loss: 0.8346 - accuracy: 0.6938 - val_loss: 0.9749 - val_accuracy: 0.6613
fig3 = plt.gcf()
plt.plot(history_4.history['accuracy'])
plt.axis(ymin=0.4,ymax=1)
plt.grid()
plt.title('Model Accuracy')
plt.ylabel('Accuracy')
plt.xlabel('Epochs')
plt.legend(['train'])
plt.show()
fig3 , ax = plt.subplots(1,2)
train_acc = history_4.history['accuracy']
train_loss = history_4.history['loss']
fig2.set_size_inches(12,4)
ax[0].plot(history_4.history['accuracy'])
ax[0].plot(history_4.history['val_accuracy'])
ax[0].set_title('Training Accuracy vs Validation Accuracy')
ax[0].set_ylabel('Accuracy')
ax[0].set_xlabel('Epoch')
ax[0].legend(['Train', 'Validation'], loc='upper left')
ax[1].plot(history_4.history['loss'])
ax[1].plot(history_4.history['val_loss'])
ax[1].set_title('Training Loss vs Validation Loss')
ax[1].set_ylabel('Loss')
ax[1].set_xlabel('Epoch')
ax[1].legend(['Train', 'Validation'], loc='upper left')
plt.show()
pip install pyvirtualdisplay
Looking in indexes: https://pypi.org/simple, https://us-python.pkg.dev/colab-wheels/public/simple/ Collecting pyvirtualdisplay Downloading PyVirtualDisplay-3.0-py3-none-any.whl (15 kB) Installing collected packages: pyvirtualdisplay Successfully installed pyvirtualdisplay-3.0
!apt update
!apt install -y xvfb python-opengl ffmpeg
from pyvirtualdisplay import Display
display = Display(visible=0, size=(1, 1))
display.start()
Get:1 http://ppa.launchpad.net/c2d4u.team/c2d4u4.0+/ubuntu focal InRelease [18.1 kB] Hit:2 http://ppa.launchpad.net/cran/libgit2/ubuntu focal InRelease Get:3 https://cloud.r-project.org/bin/linux/ubuntu focal-cran40/ InRelease [3,622 B] Hit:4 http://ppa.launchpad.net/deadsnakes/ppa/ubuntu focal InRelease Hit:5 http://ppa.launchpad.net/graphics-drivers/ppa/ubuntu focal InRelease Hit:6 https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2004/x86_64 InRelease Hit:7 http://ppa.launchpad.net/ubuntugis/ppa/ubuntu focal InRelease Get:8 http://security.ubuntu.com/ubuntu focal-security InRelease [114 kB] Hit:9 http://archive.ubuntu.com/ubuntu focal InRelease Get:10 http://archive.ubuntu.com/ubuntu focal-updates InRelease [114 kB] Get:11 http://ppa.launchpad.net/c2d4u.team/c2d4u4.0+/ubuntu focal/main Sources [2,412 kB] Get:12 http://archive.ubuntu.com/ubuntu focal-backports InRelease [108 kB] Get:13 http://ppa.launchpad.net/c2d4u.team/c2d4u4.0+/ubuntu focal/main amd64 Packages [1,143 kB] Get:14 http://security.ubuntu.com/ubuntu focal-security/main amd64 Packages [2,587 kB] Get:15 http://security.ubuntu.com/ubuntu focal-security/universe amd64 Packages [1,027 kB] Get:16 http://security.ubuntu.com/ubuntu focal-security/restricted amd64 Packages [2,060 kB] Get:17 http://archive.ubuntu.com/ubuntu focal-updates/restricted amd64 Packages [2,198 kB] Get:18 http://archive.ubuntu.com/ubuntu focal-updates/universe amd64 Packages [1,322 kB] Get:19 http://archive.ubuntu.com/ubuntu focal-updates/main amd64 Packages [3,068 kB] Fetched 16.2 MB in 5s (3,173 kB/s) Reading package lists... Done Building dependency tree Reading state information... Done 39 packages can be upgraded. Run 'apt list --upgradable' to see them. Reading package lists... Done Building dependency tree Reading state information... Done ffmpeg is already the newest version (7:4.2.7-0ubuntu0.1). The following additional packages will be installed: freeglut3 libpython2-stdlib python2 python2-minimal Suggested packages: python-tk python-numpy libgle3 python2-doc The following NEW packages will be installed: freeglut3 libpython2-stdlib python-opengl python2 python2-minimal xvfb 0 upgraded, 6 newly installed, 0 to remove and 39 not upgraded. Need to get 1,400 kB of archives. After this operation, 8,330 kB of additional disk space will be used. Get:1 http://archive.ubuntu.com/ubuntu focal/universe amd64 python2-minimal amd64 2.7.17-2ubuntu4 [27.5 kB] Get:2 http://archive.ubuntu.com/ubuntu focal/universe amd64 libpython2-stdlib amd64 2.7.17-2ubuntu4 [7,072 B] Get:3 http://archive.ubuntu.com/ubuntu focal/universe amd64 python2 amd64 2.7.17-2ubuntu4 [26.5 kB] Get:4 http://archive.ubuntu.com/ubuntu focal/universe amd64 freeglut3 amd64 2.8.1-3 [73.6 kB] Get:5 http://archive.ubuntu.com/ubuntu focal/universe amd64 python-opengl all 3.1.0+dfsg-2build1 [486 kB] Get:6 http://archive.ubuntu.com/ubuntu focal-updates/universe amd64 xvfb amd64 2:1.20.13-1ubuntu1~20.04.6 [780 kB] Fetched 1,400 kB in 1s (1,697 kB/s) Selecting previously unselected package python2-minimal. (Reading database ... 128285 files and directories currently installed.) Preparing to unpack .../python2-minimal_2.7.17-2ubuntu4_amd64.deb ... Unpacking python2-minimal (2.7.17-2ubuntu4) ... Selecting previously unselected package libpython2-stdlib:amd64. Preparing to unpack .../libpython2-stdlib_2.7.17-2ubuntu4_amd64.deb ... Unpacking libpython2-stdlib:amd64 (2.7.17-2ubuntu4) ... Setting up python2-minimal (2.7.17-2ubuntu4) ... Selecting previously unselected package python2. (Reading database ... 128314 files and directories currently installed.) Preparing to unpack .../python2_2.7.17-2ubuntu4_amd64.deb ... Unpacking python2 (2.7.17-2ubuntu4) ... Selecting previously unselected package freeglut3:amd64. Preparing to unpack .../freeglut3_2.8.1-3_amd64.deb ... Unpacking freeglut3:amd64 (2.8.1-3) ... Selecting previously unselected package python-opengl. Preparing to unpack .../python-opengl_3.1.0+dfsg-2build1_all.deb ... Unpacking python-opengl (3.1.0+dfsg-2build1) ... Selecting previously unselected package xvfb. Preparing to unpack .../xvfb_2%3a1.20.13-1ubuntu1~20.04.6_amd64.deb ... Unpacking xvfb (2:1.20.13-1ubuntu1~20.04.6) ... Setting up freeglut3:amd64 (2.8.1-3) ... Setting up libpython2-stdlib:amd64 (2.7.17-2ubuntu4) ... Setting up xvfb (2:1.20.13-1ubuntu1~20.04.6) ... Setting up python2 (2.7.17-2ubuntu4) ... Setting up python-opengl (3.1.0+dfsg-2build1) ... Processing triggers for man-db (2.9.1-1) ... Processing triggers for libc-bin (2.31-0ubuntu9.9) ...
<pyvirtualdisplay.display.Display at 0x7fb71c0d12b0>
import cv2
import numpy as np
from keras.models import load_model
from tensorflow.keras.preprocessing.image import img_to_array
from keras.models import load_model
from keras.preprocessing import image
from time import sleep
from google.colab.patches import cv2_imshow
face_classifier = cv2.CascadeClassifier('haarcascade_frontalface_default.xml')
classifier = load_model('model_1.h5')
emotion_labels = ['Angry','Disgust','Fear','Happy','Neutral', 'Sad', 'Surprise']
cap = cv2.VideoCapture('/content/test_file_1.mp4')
if not cap.isOpened():
print("Error opening video stream or file")
while True:
ret, frame = cap.read()
if not ret:
print("Error reading frame")
break
gray = cv2.cvtColor(frame,cv2.COLOR_BGR2GRAY)
faces = face_classifier.detectMultiScale(gray)
for (x,y,w,h) in faces:
cv2.rectangle(frame,(x,y),(x+w,y+h),(0,255,255),2)
roi_gray = gray[y:y+h,x:x+w]
roi_gray = cv2.resize(roi_gray,(48,48),interpolation=cv2.INTER_AREA)
if np.sum([roi_gray]) != 0:
roi = roi_gray.astype('float')/255.0
roi = img_to_array(roi)
roi = np.expand_dims(roi,axis=0)
prediction = classifier.predict(roi)[0]
label = emotion_labels[prediction.argmax()]
label_position = (x,y-10)
cv2.putText(frame,label,label_position,cv2.FONT_HERSHEY_SIMPLEX,1,(0,255,0),2)
else:
cv2.putText(frame,'No Faces',(30,80),cv2.FONT_HERSHEY_SIMPLEX,1,(0,255,0),2)
cv2_imshow(frame)
if cv2.waitKey(1) & 0xFF == ord('q'):
break
cap.release()
cv2.destroyAllWindows()
1/1 [==============================] - 0s 256ms/step
1/1 [==============================] - 0s 27ms/step
1/1 [==============================] - 0s 25ms/step
1/1 [==============================] - 0s 22ms/step
1/1 [==============================] - 0s 27ms/step
1/1 [==============================] - 0s 25ms/step
1/1 [==============================] - 0s 24ms/step
1/1 [==============================] - 0s 23ms/step
1/1 [==============================] - 0s 26ms/step
1/1 [==============================] - 0s 54ms/step
1/1 [==============================] - 0s 37ms/step
1/1 [==============================] - 0s 36ms/step
1/1 [==============================] - 0s 36ms/step 1/1 [==============================] - 0s 49ms/step
1/1 [==============================] - 0s 57ms/step
1/1 [==============================] - 0s 53ms/step 1/1 [==============================] - 0s 52ms/step
1/1 [==============================] - 0s 36ms/step
1/1 [==============================] - 0s 29ms/step 1/1 [==============================] - 0s 39ms/step
1/1 [==============================] - 0s 33ms/step 1/1 [==============================] - 0s 40ms/step
1/1 [==============================] - 0s 36ms/step
1/1 [==============================] - 0s 56ms/step
1/1 [==============================] - 0s 37ms/step
1/1 [==============================] - 0s 36ms/step
1/1 [==============================] - 0s 32ms/step
1/1 [==============================] - 0s 35ms/step
1/1 [==============================] - 0s 36ms/step
1/1 [==============================] - 0s 72ms/step 1/1 [==============================] - 0s 33ms/step
1/1 [==============================] - 0s 48ms/step 1/1 [==============================] - 0s 43ms/step
1/1 [==============================] - 0s 71ms/step
1/1 [==============================] - 0s 56ms/step
1/1 [==============================] - 0s 55ms/step
1/1 [==============================] - 0s 58ms/step
1/1 [==============================] - 0s 56ms/step
1/1 [==============================] - 0s 44ms/step
1/1 [==============================] - 0s 57ms/step
1/1 [==============================] - 0s 31ms/step
1/1 [==============================] - 0s 32ms/step
1/1 [==============================] - 0s 37ms/step
1/1 [==============================] - 0s 38ms/step
1/1 [==============================] - 0s 37ms/step
1/1 [==============================] - 0s 38ms/step
1/1 [==============================] - 0s 48ms/step
1/1 [==============================] - 0s 86ms/step
1/1 [==============================] - 0s 54ms/step
1/1 [==============================] - 0s 75ms/step
1/1 [==============================] - 0s 78ms/step
1/1 [==============================] - 0s 61ms/step
1/1 [==============================] - 0s 35ms/step
1/1 [==============================] - 0s 31ms/step
1/1 [==============================] - 0s 35ms/step
1/1 [==============================] - ETA: 0s
Summary : For the model of Emotion Detection, I tried to increase the accuracy of the model by changing hyper-parameter like learning, batch size, number of epochs, number of dense laywers, pool size. but in the end, after doing Data Augmentation I was able to get accuracy around 70%. Lastly, I tried to detect Emotions from the video by using a Pre-trained model I have created with OpenCv library which showed better results. I like to extend this project from here in "ProjectII" which will be a great start for me.